Chimeric antigen receptor (CAR)-T cell therapies have shown remarkable success against blood cancers, but tackling solid tumors has proven more challenging. Hoping to one day offer patients with advanced disease effective treatments, cancer researchers are attempting to solve this problem by exploring innovative approaches.
One approach is sherpabody-guided engineered T cells, a customizable platform that uses computer-like logic to guide CAR-T cells to specific tumor antigens, which are proteins that are overexpressed on the surface of cancer cells.
Stanford Cancer Institute member Rogelio Hernández-López, PhD, assistant professor of bioengineering and of genetics, collaborated with Wendell Lim, PhD, of UC-San Francisco, and Kalle Saksela, MD, PhD, of the University of Helsinki, on a paper describing the platform’s potential to address obstacles in CAR-T treatment for solid tumors. Hernández-López discussed how sherpabodies can both improve the ability of CAR-T cells to specifically target tumor cells and enhance their durability, allowing them to fight cancer for a longer duration.
Challenges in treating solid tumors
CAR-T cells are traditionally created by engineering T cells to recognize a specific antigen, enhancing their ability to find and attack cancer cells. To locate target antigens, conventional CAR-T cells use single-chain variable fragments (scFvs), small protein sequences derived from antibodies that sit at the tip of the cell’s engineered receptor, also known as its CAR. These scFvs are small enough to include on the receptor and retain the antibody’s ability to bind to specific antigens.
Conventional CAR-T cells are typically monospecific, meaning they bind a single antigen. Whereas most cells within a given blood cancer share the same target, solid tumors are heterogeneous, as the tumor cells vary in which antigens they express and at what levels. By binding only a single antigen, CAR-T cells will miss other cancer cells that express different antigens.
Another hurdle to the therapy’s success in solid tumors is that while cancer cells overexpress certain antigens, healthy cells can also display the same antigens at lower levels. As CAR-T cells can’t distinguish between different levels of antigen expression, they can mistakenly attack healthy cells.
Hernández-López says, “If you engineer a T cell against a protein, for example, HER2, that is highly abundant on the surface of ovarian, breast, and gastric cancers, there will be the potential for toxicity in the lungs or heart because those tissues also express HER2.”
Additionally, CAR-T cells engineered with scFvs that have high affinities, in which they bind too strongly to antigens, can undergo chronic signaling and experience CAR-T exhaustion, a well-documented complication where the cells become dysfunctional and ineffective. High-affinity CAR-T cells can also acquire the target antigen when interacting with cancer cells, causing other CAR-T cells to mistake them for cancer cells and attack.
A modular design for customized targeting
Sherpabodies aim to overcome these challenges by targeting multiple antigens and reducing chronic signaling to mitigate CAR-T exhaustion. They are derived from the SH3 domain, a protein sequence involved in cellular signaling that reliably folds into a stable 3D structure, making it a robust scaffold for engineering. Within these engineered SH3 domains is the binding site for antigens, which is the only part of the sherpabody requiring alteration to target different antigens.
Hernández-López says, “One of the big challenges of creating any binding unit is not only that they recognize the target of interest but that they are not going to stick or bind to any other proteins that you’re not trying to target.”
From a library comprising roughly one hundred billion unique amino acid sequences, his team identified over a hundred sherpabodies targeting six antigens overexpressed in solid tumors. Sherpabodies with the highest specificity in binding only to the selected antigens in the presence of similar proteins were used to engineer CAR-T cells, creating sherpabody-guided CAR (SbCAR)-T cells.
Hernández-López describes sherpabodies as a new class of antibody mimetics, as they bind antigens similarly to antibodies. He says, “Sherpabodies are a modular scaffold that can be engineered to have specificity, just like antibodies possess.”
Because the molecules are compact, it’s possible to link multiple together on a single CAR and target multiple antigens. This ability to link them together and easily customize targeting offers researchers a modular design that allows them to more efficiently recognize different antigens, as they don’t have to reengineer the whole molecule for each new target.
Using logic and synthetic circuits for improved precision
Because we know that there are not many proteins that are specific to cancer, then there is a need to create another type of logic into the T cells so that we can specifically target cancer cells."
Regarding solid tumors’ heterogeneity, Hernández-López says, “Because we know that there are not many proteins that are specific to cancer, then there is a need to create another type of logic into the T cells so that we can specifically target cancer cells.”
Addressing this need, his team designed trispecific SbCARs, which bind three different antigens, and used “OR” logic ( antigen-1 or antigen-2 or antigen-3) to cause the SbCAR-T cell to kill a cell if it expresses any one of the three antigens.
Synthetic circuits are pieces of genetic code that allow more complex logic to be programmed into engineered cells. In this study, the researchers used a circuit with “IF–THEN” logic to design an “inducible SbCAR” that controls when antigen signaling happens. If a sensor on the engineered T cell, called a synthetic Notch (synNotch) receptor, binds its target antigen, it triggers downstream signals that switch on an SbCAR, enabling antigen recognition and tumor killing. In the absence of synNotch’s target antigen, SbCAR activity remains off, letting the T cells rest between encounters and helping to limit chronic stimulation and exhaustion. The study looked at two types of inducible SbCARs: monospecific SbCARs in mouse models and bispecific SbCARs using “OR” logic in vitro.
Promising results
Both in vivo and in vitro testing with breast, ovarian, and prostate cancer cells demonstrated SbCARs’ tumor control, specificity, and persistence. In vitro, trispecific SbCARs and inducible bispecific SbCARS, both using OR logic, exhibited robust responses when encountering any one of the target antigens, illustrating the ability of sherpabodies to be used in designs with multiple antigens and logic to improve specificity when interacting with heterogeneous solid tumors.
Inducible SbCARs in vivo had better tumor control and persistence compared to always-on SbCARs, with the inducible SbCAR-T cells displaying a fitter, less-exhausted phenotype. Additionally, sherpabodies were found to have lower affinities than standard scFvs, presenting affinities more similar to natural T cell receptors. These findings suggest that switching CAR-T cells off to prevent continuous overstimulation and reducing the affinity of binding sites can mitigate exhaustion and enhance the longevity and effectiveness of CAR-T cells against solid tumors.
Of these results, Hernández-López says, “The results showed that SbCARs can have enough cytotoxic activity against cancer cells, and they provide evidence that SbCARs are modular. Each of the SH3 domains in the multispecific receptors is functional and compatible with the more complex synthetic circuits that will be needed if you are trying to engineer high specificity against the cancer cells.”
While this study serves as a proof of principle that sherpabodies can be engineered to tackle solid-tumor challenges, the next step will be to define specific diseases and antigens to test the modular platform in preclinical settings. If shown to be effective and safe in future studies, SbCARs could help overcome longstanding challenges in solid tumor treatment and expand therapeutic options for patients.